The long-term objective of this research project is to adapt microarray-based technology to measure CpG island methylation in human cancer cells. CpG islands are approximately 1kb stretches of DNA that have a high CG content, are enriched in the dinucleotide 5'-CG -3', are found at the 5' end of about 50% of all human genes, and participate in the transcriptional regulation of these genes. The cytosines in the CpG dinucleotides of CpG islands are unmethylated in normal tissue; however, CpG islands become aberrantly methylated during oncogenesis and has been linked to the transcriptional repression of the associated gene. In addition, from the limited number of CpG islands and tumors that have been analyzed to date, it appears that patterns of aberrant methylation occur in a tumor-specific and stage-specific fashion, suggesting that CpG island methylation profiles may be useful as a tumor- specific fingerprint to monitor disease activity and burden. Thus, a multiplexed assay where the cytosine methylation status of thousands of CpG islands can be determined simultaneously would be useful in the molecular profiling of human tumors, and will likely provide insights into the biology of cancer. To this end we have formed a multidisciplinary team to use human CpG island microarrays (CGI arrays) as a tool for determining CpG island methylation profiles in cancer, and from these profiles identify characteristic patterns of CpG island methylation that correlate with the tumor's clinical phenotype. The 4 integrated specific aims that follow are designed to reach our objective. 1) Construct CpG island microarrays for use in CpG island methylation analysis. 2) Optimize methylation analysis using CpG island microarrays 3) Determine CpG island methylation signatures in AML cell lines and in AML samples obtained from patients with known clinical outcome. 4) Develop and implement a public database for the dissemination and mining of the CGI array data